The data below are from a study of the effect of two treatments,
placebo (A) and active (B), on diastolic blood pressure (DBP).  Use
the randomization version of the two-sample t-test to test the null
that treatment has no effect on DBP.  That is, make all the possible
re-assignments of treatment indicators subject to the constraint that
there are four subjects in each group.  For each possible
re-assignment, calculate the conventional two-sample t statistic. 
Make the histogram of values of the t statistic over the randomization
distribution, and calculate the significance level of the test from
the probability, under this distribution, that the t statistic exceeds
its observed value from the study.  Compare this to the conventional p
value from the model-based t statistic.

Treatment/DBP
A/77
A/71
A/81
A/90
B/109
B/110
B/97
B/127